What are the benefits of “Shared-Nothing (SN)” distributed database architectures?

This post is a placeholder for a future venture into commercial benefits of Shared-Nothing (SN) distributed database architecture.

A term ‘shared nothing’ was first coined by Michael Stonebraker (University of California, Berkeley) in his 1986 database paper.

A shared nothing architecture (SN) is a type of database design where all database nodes are completely autonomous (Diagram 1), meaning that not only they do not share any of the computing resources (such as memory or storage), but they’re also completely self-reliant and independent in every other sense.

The main advantage of SN are:

  • Elimination of SPOF (single point of failure) – if a part of a system fails, it won’t stop the entire system from working
  • Non-disruptive upgrades – database updates will not interrupt access to database data
    • No need for reboots – SN does not require system as a whole to be rebooted when the update process completes.
  • Self-healing capabilities – allows a database to be repaired while still processing transactions

 

Diagram 1 (Patil, H., 2014)
distributed-database-management-system

References

Shared nothing architecture (2016) in Wikipedia. Available at: https://en.wikipedia.org/wiki/Shared_nothing_architecture (Accessed: 11 October 2016).

Single point of failure (2016) in Wikipedia. Available at: https://en.wikipedia.org/wiki/Single_point_of_failure (Accessed: 11 October 2016).

Humphries, M. (2002) Self-healing databases | mobile. Available at: http://www.geek.com/mobile/self-healing-databases-550182/ (Accessed: 11 October 2016).

Patil, H. (2014) WAISWA PHILIP. Available at: http://www.slideshare.net/hardikpatil444/distributed-database-management-system (Accessed: 11 October 2016).